0
$\begingroup$

I have time data. Different variables with some degree of correlation among them. What I would like is to pick a sample from the distribution of those who have no or low correlation with the others and somehow estimate the value for those who are correlated instead.

This being done all in a loop for a montecarlo which I will in the end aggregate.

So to recap:

1- check correlation between variables

2- for those uncorrelated plot the distribution

3- for them estimate a distribution model (like fitting the histogram)

4- run a montecarlo that for 100 times poick a random value from these distribution and estimate the value of the others since they are correlated

I have no code because I still have to start but I cannot really get how to do this in particular i know how to do point 1 and 2 but I am in trouble coding point 3 and 4.

In addition is this a good way of proceeding rather then doing a montecarlo separately for the variables even if they are correlated?

Really appreciate anybody's help

$\endgroup$
3
  • 1
    $\begingroup$ Generating correlated random variables is feasible. Take e.g. the simulation of a general multivariate Normal distribution. $\endgroup$ – Xi'an Apr 15 at 8:58
  • $\begingroup$ cool thanks for pointing. Can you provide any reference with a python implementation? $\endgroup$ – Luigi87 Apr 15 at 9:00
  • 1
    $\begingroup$ no, I do not think jumping to a code helps with understanding the fundamentals. And I do not run Python. $\endgroup$ – Xi'an Apr 15 at 9:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.